A factory acceptance test (FAT) proves that a new automated machine or system meets its specification at the builder’s facility, before it ships. A site acceptance test (SAT) repeats the critical checks after installation, on your floor, with your parts, utilities and network. Together they are how you know you are getting what you paid for, and the cheapest point to find and fix a problem.
This article is part of our industrial automation guide, which covers how manufacturers plan, specify and deploy automation. Here we go deeper on the final step of a project: proving the equipment works before you accept it.
Why acceptance testing matters
Most automation problems are cheaper to fix the earlier they are found. A wiring error, a missing interlock or a cycle time that falls short is a quick fix on the builder’s bench, where the engineers, tools and spare parts are all in the room. The same issue found after installation costs downtime, travel and a production schedule that slips while everyone waits.
Acceptance testing also turns a vague goal (“the machine should work”) into a list of measurable pass or fail criteria. That protects both sides: the buyer gets objective proof, and the builder gets a clear finish line.
FAT vs SAT: what each one proves
| Factory acceptance test (FAT) | Site acceptance test (SAT) | |
|---|---|---|
| Where | Builder’s facility | Your plant or lab, after installation |
| When | Before shipping | After installation and commissioning |
| Main question | Does the system meet the specification as built? | Does it still work here, connected to everything else? |
| Parts used | Your production samples, shipped to the builder | Live production parts and material |
| Typical checks | Function, cycle time, accuracy, safety, software, documentation | Utilities, network and MES links, interfaces to upstream and downstream equipment, operator use, sustained run |
| Who attends | Your engineering and quality leads, builder’s team | Operators, maintenance, engineering, quality, builder’s team |
| Outcome | Approval to ship, with a punch list | Final acceptance and handover, with a punch list |
Small, self-contained machines such as a benchtop counter or inspection station may combine the two into a short FAT and a quick check on arrival. Larger, integrated systems need both.
What a good FAT checks
1. Function against the specification
Every requirement in the specification gets a test step and a measurable result. If the spec says the system rejects parts outside a tolerance, the FAT runs known-good and known-bad parts and records what happens to each. Requirements that cannot be tested are a sign the specification needs another pass.
2. Throughput and cycle time
Run at production rate for long enough to see real behavior, not a single perfect cycle. Record the actual cycle time, any stalls or recoveries, and how the system behaves when parts arrive faster or slower than nominal.
3. Accuracy and repeatability
For any system that measures, inspects, counts or tests, run the same samples several times and compare results. A system that is accurate once but not repeatable will create disputes in production. For test and measurement systems, a formal gage R&R study is the standard way to quantify this.
4. Fault handling and safety
Deliberately cause the problems production will cause: jams, empty feeders, a missing part, an emergency stop, a lost network connection, a power cycle. Check that every fault is detected, reported clearly to the operator and recovered safely, and that guarding and interlocks work as designed.
5. Software, data and records
Confirm the operator screens, recipes or product changeovers, user access levels, data logging and any reports. Check that results are stored where they should be, with the right serial or lot numbers, and that software versions are recorded.
6. Documentation and spares
Drawings, manuals, the software source or backup, a spare parts list and any calibration certificates should be reviewed at FAT, not promised for later.
What a good SAT adds
- Real utilities: your power, air, lighting and environment, which can differ from the builder’s floor.
- Real integration: conveyors, upstream and downstream machines, PLCs, MES or database connections that were simulated at FAT.
- Real people: your operators running the system on their shifts, and maintenance staff walking through routine tasks.
- A sustained run: a production run long enough to expose intermittent faults, often one or more full shifts.
- Training sign-off: confirmation that operators and maintenance staff can run and support the system.
How to write acceptance criteria that work
Vague criteria cause most acceptance disputes. Each criterion should name what is measured, how, with which parts and what result passes.
| Weak criterion | Strong criterion |
|---|---|
| System runs at production speed | Average cycle time of 4.0 s or less over 500 consecutive parts, no more than 2 operator interventions |
| System detects defects | Rejects 100% of the 20 seeded defect samples across 3 runs; false reject rate below the agreed limit on 500 good parts |
| Counts are accurate | Count matches a verified reference on 10 batches of each part type, within the agreed tolerance |
| Data is logged | Each result is stored with serial number, timestamp and software version, and appears in the plant database within 5 s |
The numbers above are examples of the format, not targets. The right limits come from your process, your customer requirements and what the system was specified to do.
Common acceptance testing mistakes
- Too few real parts. Testing with a handful of perfect samples hides the variation production will bring. Send the builder a representative set, including your borderline and known-bad parts.
- Writing the test plan at the end. The FAT plan should be drafted with the specification, so every requirement is testable from day one.
- Skipping fault testing. A system that runs perfectly but recovers badly from a jam will cost more than one that is slightly slower.
- No punch list discipline. Record every open item with an owner and a date, and agree which items block shipment and which can be closed on site.
- The wrong people at FAT. Bring the people who will live with the system: a lead operator and a maintenance technician see problems engineers miss.
For the person signing off: what acceptance testing protects
For an operations leader, acceptance testing is risk control on a capital purchase. A clear FAT and SAT plan means payment milestones are tied to proven performance, the system is fit for production on day one, and your team has the training, documentation and spares to keep it running. Ask any automation partner to show you a sample FAT plan before you sign.
Planning an automation project?
Sciotex designs, programs, builds and supports custom automation in the US, with free sample-part testing up front and acceptance testing built into every project plan. No subscription or support contract required.
Frequently asked questions
A factory acceptance test (FAT) is a structured test of a new machine or automated system at the builder’s facility, before shipping, to prove it meets its specification using the buyer’s parts and agreed pass or fail criteria.
A FAT proves the system meets its specification at the builder’s facility. A site acceptance test (SAT) repeats the critical checks after installation, with your utilities, network, connected equipment, operators and live production.
Your engineering and quality leads, plus the people who will run and maintain the system, such as a lead operator and a maintenance technician, along with the builder’s engineers.
It depends on the system. A small benchtop machine may need a few hours; a large integrated line can take several days. The plan should include a sustained run at production rate, not just single cycles.
Failures go on a punch list with an owner and a due date. The buyer and builder agree which items must be fixed and retested before shipment and which can be closed during SAT.